74 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



Copper, while diamagnetic as metal or in cuprous compounds, is 

 paramagnetic in cupric compounds, is attributable to this tauto- 

 merism also. 



In monovalent Cu, Ag, Au, and in the salts of Zn, Cd, Hg, there 

 are 8 free magnetons left, and the tautomerism 8 ^ y is still pos- 

 sible : therefore we expect, and find, diamagnetism. But in bivalent 

 Cu, where only 7 are left, this tautomerism is no longer possible, and 

 the salts are strongly paramagnetic, as this theory would predict. 

 Another prediction — that AuBr (8 ^= y) should be diamagnetic, and 

 AuBr 3 (6) paramagnetic — is not so successful, for both are dia- 

 magnetic : but the obvious refuge from the difficulty will suggest 

 itself. Compounds of bivalent or trivalent Silver are of course not 

 available for comparison. 



The complex salts of these metals were also studied, in the hope of 

 getting results analogous to the well-known relations for Fe and 

 Co (F). Bivalent Cu was obviously the best point of attack, 

 but the most stable complex cupric salt obtainable seems to be 

 Cu(NH 3 ) 4 S0 4 • H 2 0, and this is still very paramagnetic: a cupri- 

 cyanide (K 2 Cu(CN) 4 ), if it were stable, might be expected to show 

 a much diminished paramagnetism, just as ferri-cyanides do. The 

 complex cyanides derived from salts which are diamagnetic already, 

 e. g., those of Cu 1 (8^ 7), Ag 1 (8 +7 y), Zn 11 (8 -^ y), are all 

 diamagnetic, although it is hard to see how, in a small complex 

 molecule like KAg(CN),, all the 8 free magnetons of the monovalent 

 Ag atom can be involved. However, the effect we should be inclined 

 to look for in such cases — a paramagnetism — has been observed in 

 one compound at least. Pascal found that the salt K 2 HgI 4 in solution 

 is paramagnetic ; so it seems that not all of the free magnetons of 

 bivalent mercury, Hg 11 (8 ^Fy), are involved in this case. 



§22. The Dependence of Magnetism upon Temperature and 



Physical State 



In the preceding sections (esp. §19), the influence of neighboring 

 atoms and molecules on one another's magnetism has been continu- 

 ally spoken of, and it has been brought out in a general way that 

 this may be expected to diminish a resultant paramagnetism or 

 increase a resultant diamagnetism. A summary of the experimental 

 evidence on this point will now be given ; and in considering this, it 

 should be remembered that the influence of one atom or molecule 

 upon another becomes diminished as the temperature rises. 



